Abstract
Linear parameter estimation in spectrophotometry (LPES), least squares calculation to extract the maximum possible information from the spectra, is simulated by means of a computer. In simulation, the standard and sample spectra are synthesized with Gauss function. After adding random noise to the spectra, regression analysis is done to estimate the linear parameter which represent the ratio of standard and sample spectra. Some conditions related to the measurement, such as noise magnitude, reproducibility of wavelength, and spectra characteristics, such as band width at half height, proximity of spectra are varied and the effect of those conditions to the precision of measurement are investigated. In the case of single component system, LPES improves the precision of measurement, and the errors due to noise and impurity spectra are reduced as compared with usual single wavelength measurement. In multicomponent system, LPES can be used to determine the composition of mixture conveniently, but some limitations to the system, such as proximity of spectra, absorbance ratio, are found.
